-
Notifications
You must be signed in to change notification settings - Fork 9
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Current Build Windows Errors #61
Comments
OK, the first git error is because
|
is failing because there's no earlier revision:
|
This sucks. |
We wouldn't get paid the big bucks for this if it were easy. |
:D. Hmmm many of these errors I've no idea about... maybe just to confirm you're on latest autopep8 and docformatter. I guess the branch still existing means the clean up (which deletes the .hg directory each time Can you try manually deleting the temp directory completely before running tests? |
Autopep8 and docformatter are current to what's on PyPI. Deleted everything in temp, same error. |
drat... |
...wait when you say deleted everything in temp, did you try deleting temp? Those are dot files so may be hidden (.hg and .git directories). :s solved the skipping bzr thing (the skiptest was in check rather than setUp). |
OK, deleting temp itself fixes those two errors about the branch. Others still present. |
Can you pdb into test_earlier_revision (test_pep8radius.TestRadiusHg) and say what's rev1 and rev2? Aaargh Windows. It's weird cos the others reads as if it's tracking all the files in the temp directory when I would expect it (slash, on Windows/Mac) it only tracks those which we've explicitly added. At least that's part of the issue. :tears: |
http://stackoverflow.com/questions/4950637/setting-breakpoints-with-nosetests-pdb-option Gonna tattoo this onto the back of my hand. And perhaps one day I'll find out why replacing |
Here's an edited log as I step through the code. Note that
|
Oh, dear me, what a very funny thing to be potentially be the cause of so much trouble. Please remember, violence is not a solution, even to dimwitted Windows peeps who didn't bother to actually set up Hg.
Step through it a bit.
I think there winds up being a
And so the actual hg error doesn't git captured by |
Bumping. As an idea, both to fix this and for ease of maintenance, maybe we have one layer of abstraction too many for a project of this size? Maybe just propagating the error another level up on that last code sample makes sense. I defer to your judgment on this - my attempts at architecture are wanting. (There's a great comic I can't find on this - programmer's vision vs. implementation, with a house).
|
Never quite got how nosetests plays with capturing stdout, the thing is we already capture it in the test (so we can integration test!). Maybe it would be easier just not to integration test (but the code is already there... except for Windows). I still don't get how hg can pass on mac/linux, but fail on windows. I think it's an hg issue (behaviour differs on platforms)... |
Check the long post before my last one. Hg problem was that user name wasn't set, but that error gets swallowed before it propogates to the output. If you have EDIT: Hg works once the user name is set. I imagine you'd have the same trouble on OSX/Linux if you delete the user name from the config file (i.e., how a fresh install looks). |
So the whole ter thing was a red herring? (was to do with your user config?) nose -s thing is often broken... I is often fixed by just using pytest! |
It's a red herring, but take a look at the long post I have about how the useful error message from Hg gets wiped out before it percolates all the way up. I think we need to make sure the error messages get out and don't get wiped clean and become an unexplained failure.
Instead of just returning false, it make sense to also pass back the errors as |
This are/should be reraised in main, at least the CalledProcessError ? |
Delete your Hg username for your Hg config and run nosetests and see what you get. What should happen is what happens in the first post of this thread - you get back a bunch of errors that make it look like the comparison to the test is failing; if you dig a little deeper, it looks like Hg is returning an empty string for some reason, but if you finally (and belatedly on my part, sorry) go all the way to the bottom of the stack trace you find Hg is returning an error about the username, but it looks like the Anyway, I'd recommend trying it on your build (the delete Hg username) and see if that error makes it out of nosetests or if something else percolates up. |
I'm ok with the tests failing like that if not properly configured. More On 14 August 2014 12:13, Shayne Hodge notifications@github.com wrote:
|
Should the first test be skipping if no Bzr is installed?
The text was updated successfully, but these errors were encountered: